Starter
For teams with one clear use case and modest volume.
Concrete plans with the LLM cost included — no hidden token bills. Shape your setup, we'll show you the right plan.
A base fee, an included turn allowance, and a clear price per additional turn — LLM cost already in. We route internally to the cheapest model that solves the task.
For teams with one clear use case and modest volume.
For production rollouts with steady traffic and multiple use cases — voice widget included.
For high volumes and complex setups with multi-agent graphs and voice widget.
A turn is one complete user question and the agent's response. The turn price already includes the LLM cost — you pay one price per turn, no separate token bill. Internally we route to the cheapest model that solves the task (smaller models for simple answers, larger ones only where needed).
The platform is fully self-serve — you can configure everything yourself. If you'd rather go live faster, book one of our setup packages: we build the knowledge bases, model the personas, and wire the integrations for you.
Web chat widget, 1 knowledge base (up to ~500 URLs), standard persona, 1 webhook tool, custom branding.
Web chat and voice widget, 2–3 knowledge bases, custom persona, 2–4 webhook tools, single sign-on.
Multi-agent graphs, structured knowledge bases with bespoke modelling, complex integrations, Azure OpenAI EU, voice and telephony, guardrails hardening.
When usage swings hard — lots of document uploads, image processing, long RAG contexts, or deliberately more expensive models (Opus, reasoning) — a flat per-turn price stops being a fair deal. With token passthrough you pay a base fee plus your included LLM budget; usage above is billed at the actual token cost plus a moderate markup.
Base fee identical to the standard plan. On your invoice you see two clearly separated lines: platform base fee and actual token cost with markup.
If you'd like to operate ChatFlow as a multi-tenant platform for subsidiaries, business units, or end customers — or as an agency reselling under your own brand — take a look at our partner programs: enterprise license and white-label partner with revenue share.
A turn is one complete user question and the agent's response. The turn price already includes the LLM cost — no separate token bill.
Each additional turn is billed at the per-turn price shown in your plan (€ 0.12 / 0.09 / 0.06). No surprises, no throttling — the plan just keeps running.
Tiered is right for predominantly text conversations with predictable load. Token passthrough makes sense as soon as you offer document or image uploads, or deliberately default to more expensive models — at that point a flat fee can no longer absorb the variance.
Yes, on request. From roughly 5,000 conversations per month, or for pure voice/telephony use cases, we can switch you to usage-based pricing (per resolution, per minute, or per outcome) tailored to your scenario. Just bring it up in the intro call.
+ € 149 / month base fee and + € 0.02 per turn surcharge. You get EU data residency and a full data processing agreement.
The voice widget (realtime/WebRTC) is included from the Growth plan onward and can be activated directly. Telephony over SIP (FreeSWITCH, Twilio, Vonage) is a separate add-on on every plan, because it requires provider integration and trunk configuration. Conversation minutes are billed by usage.
Yes. Starter is built for that — one use case, one channel, live fast, then grow. Plan changes are possible anytime, prorated to the month.
Self-hosting and multi-tenant operation aren't part of the plans above — they're covered by our enterprise license or white-label partner program. Details on the partner programs page.